ABSTRACT
The previous chapters have showcased a number of signal processing approaches for gas turbine diagnostics. In general, the filters have become more complex and powerful from Chapter 2 to Chapter 6. The issue of nonGaussian noise remains an important aspect for filters developed for fault detection. For example, Yoshida [71] points out that non-Gaussian noise occurs in health signals because damage tends to be concentrated in a specific part of the structure. He used a Monte Carlo filter to address the issue of non-Gaussian noise in structural damage detection for structures following earthquakes. However, the computer time requirements for such a filtering method can be very large.
